AI and Blood Tests Could Catch Diabetes Years Earlier Than Current Methods

Two groundbreaking studies suggest that artificial intelligence and advanced blood testing could revolutionize how doctors predict type 2 diabetes risk, potentially catching the disease years before current methods. With about 1 in 9 adults worldwide living with diabetes—over 90% having type 2—early detection could be a game-changer for preventing complications.

How Does the AI Platform Work?

Researchers developed an artificial intelligence platform called GluFormer that analyzes continuous glucose monitoring (CGM) data to predict diabetes risk. The AI was trained using more than 10 million glucose measurements from almost 11,000 adults, most of whom didn't have diabetes at the time.

In a study with 580 participants followed for 11 years, GluFormer significantly outperformed the standard hemoglobin A1c (HbA1c) test, which measures average blood sugar levels over three months. The results were striking: 66% of participants categorized as highest-risk by the AI platform developed diabetes, while only 7% of those in the lowest-risk category did.

What Makes Blood Molecule Testing Different?

The second breakthrough involves identifying small molecules called metabolites in blood samples that could predict future diabetes risk beyond traditional factors like age, weight, and family history. "Metabolites are small molecules found in our blood that are produced during our bodies' daily activities, such as natural biological processes, to maintain function, when we eat, store or use energy, and respond to everyday activities like exercise," explained Dr. Jun Li, assistant professor of medicine at Mass General Brigham.

These metabolites act as chemical "footprints" that reflect how well the body's metabolism is working at any given moment, potentially revealing diabetes risk before symptoms appear.

Why Current Testing Falls Short

Type 2 diabetes presents unique diagnostic challenges that make early detection difficult with current methods:

• Slow Development: The disease develops gradually, and by diagnosis time, damage to the heart, kidneys, or blood vessels may have already begun
• Hidden Symptoms: Symptoms take a long time to develop or may not show themselves at all, making detection challenging
• Limited Risk Factors: Current evaluation tools rely mainly on age, body weight, family history, and blood sugar levels, which don't capture underlying biological changes leading to diabetes

"Current risk evaluation tools rely largely on risk factors such as age, body weight, family history, and blood sugar levels. Although helpful, these measures do not capture the underlying biological changes that lead to diabetes, and many people who eventually develop the disease are not flagged as high risk early enough," said Dr. Jun Li.

The GluFormer platform also showed promise for cardiovascular risk prediction. Among participants categorized as high-risk, 69% died from heart-related conditions during the study period, while no deaths occurred in the lowest-risk group.

What Are the Next Steps?

While these advances show promise, implementation challenges remain. Dr. David Cutler, a family medicine physician at Providence Saint John's Health Center who wasn't involved in the studies, noted important considerations: "The question remains whether better risk prediction with the GluFormer model of continuous glucose monitoring data will lead to better outcomes. Once patients are told they are more likely to develop diabetes or have a heart attack, will they take the medication, change their behaviors, and undergo procedures which will treat diabetes and prevent heart attacks and strokes?."

Cost considerations for continuous glucose monitoring devices and data interpretation, along with the time needed to transition healthcare providers to new technologies, could affect widespread adoption. Historical precedent suggests it often takes a decade or more for proven beneficial technologies to become routine practice.

These research developments offer hope for catching diabetes in its earliest stages, when interventions might be most effective at preventing the disease's progression and serious complications like diabetic neuropathy and cardiovascular disease.